13 May
2007
13 May
'07
1:48 a.m.
Message: 8
Date: Sat, 12 May 2007 11:49:02 +0100
From: Paul Williams <paul at frixxon.co.uk>
Subject: Re: Manuals being scanned
To: General Discussion: On-Topic and Off-Topic Posts
<cctalk at classiccmp.org>
Message-ID: <46459B9E.4040706 at frixxon.co.uk>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Al Kossow wrote:
>
> They are some of the ugliest, bloated pdf's I've ever seen.
> What are you doing to them to make them look so BAD.
This is Adobe Capture at work. I had a look at the
CP/M manual to see if
the OCRed text was overlaid on an original full page image beneath, but
unfortunately, the 256-page document consists of 13441 image fragments
that couldn't be OCRed. There are no images where text has been OCRed.
--
Paul
Paul -
in all kindness (I've been there, done that)
Stop scanning a bit and consider:
1) your current technique is not good!
a) your current files are indeed much too large
by maybe a factor of 6 or 7
b) your current files contain OCR errors - with no warning
to users, who tend to think .pdf means a valid image.
Two easily found errors are on "your"
http://www.cse.uta.edu/TheMuseum at
CSE/Manuals%20Scanned/CDC/Control%20Data-Cyber%2070%20Computer%20Systems%20Models%2072,73,74,6000%20Computer%20Systems.PDF
on page 3-7 or page 19 in the file
in HTML, several variable which should be
l<sub>i</SUB>
have been mis OCRed and are shown as
I,
Adobe blunders such as the above caused me to abandon
Adobe for several years, until I met an Adobe employee
who straightened me out.
I now use Adobe Acrobat Professional as scanning input control,
(not optimum, but it works OK)
- trim off the images of paper holes with "Crop Pages"
- check the "Recognize Text using OCR"
which places Adobe's interpretation "behind" the image
(for direct searching)
- and have figured how to make "Reduce File Size" work
Another thing to consider is that GOOGLE OCRs .pdf
files that it spiders, highlighting search hits :-))
To some extent, your Adobe OCR is bypassed by GOOGLE.
So - I'm back "PDF"ing,
http://www.ed-thelen.org/comp-hist/on-line-docs.html
http://www.ed-thelen.org/#h-documents
etc.
For serious work, I suggest you look at Al Kossow's suggestions at
http://www.bitsavers.org/
"Keep Smiling" ;-))
Ed Thelen
5:27 p.m.
"Ed Thelen" wrote:
Another thing to consider is that GOOGLE OCRs .pdf
files that it spiders, highlighting search hits :-))
really?
can you provide any links to back that up? (I mean, I believe you, but I've
never read anywhere that that was happening and I'm curious about ti)
-brad
6:36 p.m.
On May 13, 2007, at 5:27 PM, Brad Parker wrote:
Umm. Have you ever googled for something and had it turn up PDF
files? I'd say fully a third of them that I run across contain
images of pages, not text.
-Dave
--
Dave McGuire
Port Charlotte, FL
Another thing
to consider is that GOOGLE OCRs .pdf
files that it spiders, highlighting search hits :-))
really?
can you provide any links to back that up? (I mean, I believe you,
but I've
never read anywhere that that was happening and I'm curious about ti) 
6:44 p.m.
In article <AEEB9762-D713-4BD6-B993-5D96A7E5BD49 at neurotica.com>,
Dave McGuire <mcguire at neurotica.com> writes:
Umm. Have you ever googled for something and had
it turn up PDF
files? I'd say fully a third of them that I run across contain
images of pages, not text.
How do you know it doesn't contain text? As described earlier in this
thread, the PDF file can be constructed with a "hidden" text layer
that is OCR'ed from the images so that text searches work. I bet that
Google just searches embedded text because they never find stuff on
bitsavers when you search for text.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
14 May
14 May
2:09 a.m.
On May 13, 2007, at 6:44 PM, Richard wrote:
Because in several instances in which I've downloaded the
referenced PDF file and torn it apart (to re-compress, trim out cover
pages that weren't present in the original document, etc) they've
contained only images.
Umm. Have
you ever googled for something and had it turn up PDF
files? I'd say fully a third of them that I run across contain
images of pages, not text.
How do you know it doesn't contain text? As described earlier in this
thread, the PDF file can be constructed with a "hidden" text layer
that is OCR'ed from the images so that text searches work.
Yup. That's a damn nice feature of the PDF format if you ask me.
I bet that
Google just searches embedded text because they never find stuff on
bitsavers when you search for text.
I'm sure it does that too.
-Dave
--
Dave McGuire
Port Charlotte, FL
2:10 p.m.
On Monday 14 May 2007 02:09, Dave McGuire wrote:
On May 13, 2007, at 6:44 PM, Richard wrote:
Because in several instances in which I've downloaded the
referenced PDF file and torn it apart (to re-compress, trim out cover
pages that weren't present in the original document, etc) they've
contained only images.
What tools are you using to do this? I have a number of datasheets that I
haven't put online yet because they're "branded" with other sites
feeling the
need to put a "page" in there saying where they've been downloaded from,
for
example, or other similar nonsense.
Something that runs under linux would be preferred.
--
Member of the toughest, meanest, deadliest, most unrelenting -- and
ablest -- form of life in this section of space, ?a critter that can
be killed but can't be tamed. ?--Robert A. Heinlein, "The Puppet Masters"
-
Information is more dangerous than cannon to a society ruled by lies. --James
M Dakin
Umm.
Have you ever googled for something and had it turn up PDF
files? I'd say fully a third of them that I run across contain
images of pages, not text.
How do you know it doesn't contain text?
5:21 p.m.
On May 14, 2007, at 2:10 PM, Roy J. Tellason wrote:
Because in several instances in which I've downloaded the
referenced PDF file and torn it apart (to re-compress, trim out cover
pages that weren't present in the original document, etc) they've
contained only images.
What tools are you using to do this? I have a number of datasheets
that I
haven't put online yet because they're "branded" with other sites
feeling the
need to put a "page" in there saying where they've been downloaded
from, for
example, or other similar nonsense.
Something that runs under linux would be preferred.
I use PDFLab under MacOS X. I've not yet found anything portable
that does the job...that's not to say that it doesn't exist, just
that I haven't found it.
-Dave
--
Dave McGuire
Port Charlotte, FL
Umm. Have
you ever googled for something and had it turn up PDF
files? I'd say fully a third of them that I run across contain
images of pages, not text.
How do you know it doesn't contain text? 
6:26 p.m.
On Mon, May 14, 2007 at 05:21:11PM -0400, Dave McGuire wrote:
I use PDFLab under MacOS X. I've not yet found
anything portable
that does the job...that's not to say that it doesn't exist, just
that I haven't found it.
There's a fairly nice command-line based set of Java tools in
multivalent that allows you to process PDFs in various different ways.
I use it mostly to compress and decompress PDFs, but it also allows you
to extract text, reorder pages, split and merge documents and so on.
About the only thing I'd like it to do that it doesn't is to extract
image resources from a PDF.
It can be found at
http://multivalent.sourceforge.net/
Cheers,
Brian.
15 May
15 May
4:30 a.m.
On May 14, 2007, at 6:26 PM, Brian Foley wrote:
Ahhhh, that will be VERY useful...Thanks for the pointer!
-Dave
--
Dave McGuire
Port Charlotte, FL
I use PDFLab
under MacOS X. I've not yet found anything portable
that does the job...that's not to say that it doesn't exist, just
that I haven't found it.
There's a fairly nice command-line based set of Java tools in
multivalent that allows you to process PDFs in various different ways.
I use it mostly to compress and decompress PDFs, but it also allows
you
to extract text, reorder pages, split and merge documents and so on.
About the only thing I'd like it to do that it doesn't is to extract
image resources from a PDF.
It can be found at
http://multivalent.sourceforge.net/ 
11:08 p.m.
What tools are you using to do this? I have a number
of datasheets that I
haven't put online yet because they're "branded" with other sites
feeling
the
need to put a "page" in there saying where they've been downloaded from,
for
example, or other similar nonsense.
I don't think these pages NEED to be chopped. But if you are up to it,
use primoPDF printer driver/pdf creator and print the datasheet from page 2
to end. Presto! There goes the annoying page ;o)
13 May
13 May
8:33 p.m.
On Sunday 13 May 2007 17:27, Brad Parker wrote:
"Ed Thelen" wrote:
Um, how about any number of google searches that I've done that bring up
links to pdf files?
--
Member of the toughest, meanest, deadliest, most unrelenting -- and
ablest -- form of life in this section of space, ?a critter that can
be killed but can't be tamed. ?--Robert A. Heinlein, "The Puppet Masters"
-
Information is more dangerous than cannon to a society ruled by lies. --James
M Dakin
Another thing to consider is that GOOGLE OCRs
.pdf
files that it spiders, highlighting search hits :-))
really?
can you provide any links to back that up? (I mean, I believe you, but
I've never read anywhere that that was happening and I'm curious about ti)
9:37 p.m.
> > Another thing to consider is that GOOGLE
OCRs .pdf
> > files that it spiders, highlighting search hits :-))
> can you provide any links to back that up? (I
mean, I believe you, but
> I've never read anywhere that that was happening and I'm curious about ti)
Um, how about any number of google searches that
I've done that bring up
links to pdf files?
Doesn't mean they're OCRed. Run a 'strings' on the pdf -- the text is
there.
--
------------------------------------ personal: http://www.cameronkaiser.com/ --
Cameron Kaiser * Floodgap Systems * www.floodgap.com * ckaiser at floodgap.com
-- LDA #<TXT:LDY #>TXT:JMP $AB1E:TXT .asc "just another c64 hacker":.byt
0d00 -
9:49 p.m.
On Sunday 13 May 2007 21:37, Cameron Kaiser wrote:
Um, how about any number of google searches that I've done that bring up
links to pdf files?
Doesn't mean they're OCRed. Run a 'strings' on the pdf -- the text is
there.
Um, yup...
!Z#OM'<k$.<r%$$)0Q$]Y18tR!nO3mTZOsET:TN.1fhLkg6hr,f/6"q#[:Oi+cpuq
i,is=:K(n`ou&\BA+Hb-)(&o8lin7qoh/K]QebL7rnbplKT?Vc84et4K<I5,XDDc.
E]TtR[]0^60<"4'/EQMo\U5Saj7SXCJkKiNA;UQIQU8XFQ'N<ahNb'o9''`qe4K:=
B0g#\8?g+`<R8I]/XO=3=/7<b]N`'lmG6?V7?,#1'G!i==\mu2*h at YBpcORH%uR];
+l4"_/Vaa>*K;Gjm*Nq$_Gq.?_/TC>34_o2H-J)TH`D(Gn"tTc!E#.FS+u*kL;Dcb
0h@\8QP[]]js*'5m@[9\EAYI8X`LQ?c6/u57m=mko2,Q/<&qioVsuk at pV46IoEG??
)t,,kbAn-LBC*MCRqY(h!0kjLR/~>
endstream
endobj
7 0 obj
<</Filter /FlateDecode
/Length 291
stream
?J?@
???l<JU?&.?`=
J?ii+?
?JE???????|3{
??|??
w`?3
DEb$?~
1??%Q??
?(????=??=?|?
???X#
?xjL??:J?GV???P???
??fjQ
e?}i???r#
C?J?<?s
Et?,??Pq)
????
Nope, I don't think any of that looks particularly useful...
--
Member of the toughest, meanest, deadliest, most unrelenting -- and
ablest -- form of life in this section of space, ?a critter that can
be killed but can't be tamed. ?--Robert A. Heinlein, "The Puppet Masters"
-
Information is more dangerous than cannon to a society ruled by lies. --James
M Dakin
Another thing
to consider is that GOOGLE OCRs .pdf
files that it spiders, highlighting search hits :-))
can you provide any links to back that up? (I mean, I believe you, but
I've never read anywhere that that was happening and I'm curious about
ti)
10:38 p.m.
Um, how about any number of google searches that
I've done that bring up
links to pdf files?
Doesn't mean they're OCRed. Run a 'strings' on the pdf -- the text is
there. 
11 p.m.
Cameron Kaiser wrote:
Doesn't mean they're OCRed. Run a 'strings' on the pdf -- the text is
there.
I suspect that the <stream> encoding in the PDF's probably needs to be
deciphered to
see the text. I also suspect there may be a variant that has plain
text. Note that both tiff
and postscript can be encapsulated in pdf's with very little
modification and in the case
of some postscript files, the plain text will be visible and ascii
readable, and in others
it will not (such as EPS).
The original documents in the thread seem to have been touched and perhaps
OCR'ed by Acrobat 6. I used Acrobat 6 to capture documents, and found a
similar problem when I used it, and had to turn off the "document" capture
and change to picture capture options. When I did that, the size increased
dramatically, so I ended up abandoning that format.
After a lot of experimentation, I frankly have not found much to change
about
how Bitsavers is encoded and stored. At a future time someone may OCR the
archive, but right now I don't think that the resources exist to scan
and store
the documents in a fashion that current OCR software will work. But the
black
and white images work for humans, so sometime soneone may get an OCR
package, modify it, and make a run at the database.
Jim
14 May
14 May
2:11 a.m.
On May 13, 2007, at 9:37 PM, Cameron Kaiser wrote:
UNLESS the PDF file simply consists of scanned images of pages,
which is frequently the case.
-Dave
--
Dave McGuire
Port Charlotte, FL
>>
Another thing to consider is that GOOGLE OCRs .pdf
>> files that it spiders, highlighting search hits :-))
> can you provide any links to back that up?
(I mean, I believe
> you, but
> I've never read anywhere that that was happening and I'm curious
> about ti)
Um, how about any number of google searches that
I've done that
bring up
links to pdf files?
Doesn't mean they're OCRed. Run a 'strings' on the pdf -- the text
is there.
6272
days inactive
6275
days old
16 comments
10 participants
participants (10)
-
alexandre-listas＠e-secure.com.br -
bfoley＠dcs.warwick.ac.uk -
brad＠heeltoe.com -
ed＠ed-thelen.org -
jwstephens＠msm.umr.edu -
legalize＠xmission.com -
mcguire＠neurotica.com -
paul＠frixxon.co.uk -
rtellason＠verizon.net -
spectre＠floodgap.com